True, but what if the sensors are not working as Louis suggested and OP is in the beginning of seeing the CPU just slowly frying itself until system instabilities do start to occur?
Not that this cannot happen, but it is relatively rare.
What it comes down to is that, unless you're willing to pull the thing apart, replace thermal paste, etc., and put it back together you're not going to see any change (unless something's clogged/dirty, and I presume whatever cleaning that can be done has been done). You may not see any change afterward, either.
I have yet to see a CPU not give scads of warning signs when failure is in the cards, whether that failure is imminent or a bit further off. If I've got a system that's behaving normally with temperature spikes secondary to heavy-duty processing - and in my experience modern games involve some very heavy duty processing on a cyclic basis - I don't worry about it.
Some of that comes, I guess, from having my very earliest laptops using processors that were known to "run hot" (and on my now ancient Acer that was an understatement) straight out of the box. It's still running (but I'm not using it practically at all these days - it's an XP era box) many, many years later.
I really wish that manufacturers would routinely include a normal "idle temperature range" along with a "normal stressed temperature range," and always report what the Tcritical temperature is for a given CPU/APU/GPU. I have been astounded on several occasions at just how big a spread there is between what is reported as maximum normal operating temperature and Tcritical, where the machine will shut itself down to prevent destruction. There's usually a generous amount of breathing room for transient spikes, but sometimes the breathing room is so great that I find the "maximum normal" temperature to be a joke.